On the Accuracy in Modeling the Statistical Distribution of Random Telegraph Noise Amplitude

The power consumption of digital circuits is proportional to the square operation voltage and demand for low reduces towards threshold MOSFETs. A weak signal makes vulnerable noise optimization circuit design requires modelling noise. Random Telegraph Noise (RTN) dominant modern CMOS technologies Monte Carlo has been used assess its impact on circuits. This statistical distributions RTN amplitude three different were proposed by early works: Lognormal, Exponential, Gumbel distributions. They give substantially predictions agreement not reached which distribution should be used, calling accuracy into questions. objective this work these explore other better accuracy. novel criterion selecting distributions, a monotonic reduction errors with increasing number traps. existing do meet thirteen are explored. It found that Generalized Extreme Value (GEV) lowest error new criterion. Moreover, reduce errors, works bimodal Lognormal Exponential have more fitting parameters. Their however, still higher than those monomodal GEV distribution. long tail predicts worse impact. highlights uncertainty in predicting models.